Unverified50% confidenceFactExact time
Google DeepMind在2024年提出工具使用审计（Tool Use Auditing）框架，通过独立验证器在工具调用执行前进行安全检查
1
Sources
50%
Confidence
Long-term
Relevance
8/13/2026
First Seen
Sources
Related Claims
VerifiedGoogle DeepMind released the "Frontier Safety Framework."75% similarUnverifiedGoogle DeepMind 的 CaMeL 框架（2025年）提出通过双 LLM 架构隔离指令执行与内容解析的系统性防御方案73% similarUnverifiedGoogle DeepMind在2023年提出了能力弹性（Capability Elicitation）框架，强调模型的真实行为边界只有在特定触发条件下才会暴露72% similarUnverifiedGoogle DeepMind, Microsoft, and xAI have agreed to submit powerful new models to CAISI under the U.S. Department of Commerce for safety review before release.70% similarUnverified谷歌为治理Flaky Test开发了内部工具来自动标记和隔离Flaky Test，并建立了专门的治理流程69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/742059API
curl https://kongchang.com/api/v1/knowledge/claims/742059MCP
get_claim(id=742059)